📙 Language Model

=== a 📊 Probability distribution over words or word sequences==

How?

Unigrams

• Only use word features and use all words → 🤦‍♂️ Naive Bayes

N-Grams (2+)

1. Compute the conditional 🎲 Probability for all words w of W:

   • $P(w|W_1..W_{n-1})$

2. Choose the word combination with the highest conditional probability (= most likely follow up word)

🔗 Links

🔮 Word Prediction